Assessment of the Immediate Impacts of the 2013•Fi2014 Drought On
Western North American Naturalist
Volume 75 Number 2 Article 1
8-21-2015
Assessment of the immediate impacts of the 2013–2014 drought on ecosystems of the California Central Coast
Christopher Potter Ames Research Center, Moffett Field, CA, [email protected]
Follow this and additional works at: https://scholarsarchive.byu.edu/wnan
Part of the Anatomy Commons, Botany Commons, Physiology Commons, and the Zoology Commons
Recommended Citation Potter, Christopher (2015) "Assessment of the immediate impacts of the 2013–2014 drought on ecosystems of the California Central Coast," Western North American Naturalist: Vol. 75 : No. 2 , Article 1. Available at: https://scholarsarchive.byu.edu/wnan/vol75/iss2/1
This Article is brought to you for free and open access by the Western North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Western North American Naturalist by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], [email protected]. Western North American Naturalist 75(2), © 2015, pp. 129–145
ASSESSMENT OF THE IMMEDIATE IMPACTS OF THE 2013–2014 DROUGHT ON ECOSYSTEMS OF THE CALIFORNIA CENTRAL COAST
Christopher Potter1
ABSTRACT.—A methodology was developed to assess the impacts of the historic 2013–2014 drought on ecosystems of California’s Central Coast region, using a combination of satellite image analysis and in situ measurements of soil mois- ture in predominant vegetation types of the region. According to differences in Landsat drought indices for plant water stress and vegetation green cover (NDWI and NDVI, respectively), the geographic areas within the study region that were most severely impacted by the 2013 drought were the inland Carmel Valley in northern Monterey County and the coastal zones around San Simeon Point and Cambria in northern San Luis Obispo County. An expanded area of severe vegetation moisture stress, generally indicated by relative-differenced NDWI values of >400, was detected by May 2014 in both of these same geographic areas. For more detailed examination of drought impacts, the entire study region was separated into the 3 predominant vegetation types (grasslands, shrublands, and forests) to examine changes in Land- sat NDWI and NDVI in the context of differing plant community responses to severe drought. Results confirmed that higher overall drought stress in 2013 and 2014 was detected in grasslands, compared to shrublands and forests, in both years at all elevations and slopes >200 m and >5%, respectively.
RESUMEN.—Se desarrolló una metodología para evaluar el impacto de la sequía histórica del 2013-2014 en los ecosistemas de la región de la costa central de California, utilizando una combinación de análisis de imágenes satelitales y mediciones in situ de la humedad del suelo en los tipos de vegetación predominantes de la región. De acuerdo a las diferencias en los índices de sequía Landsat para el estrés hídrico de las plantas y la cubierta vegetal (NDWI y NDVI, respectivamente), las áreas geográficas dentro de la región de estudio que fueron más gravemente afectadas por la sequía del 2013 fueron Carmel Valley en el norte del Condado de Monterey, y las zonas costeras alrededor de San Simeon Point y Cambria en el norte de San Luis Obispo County. Un área extensa de estrés de humedad severo en la vegetación, en general indicado por la diferencia relativa de valores NDWI de más de 400, se detectó en mayo del 2014 en estas dos mismas áreas geográficas. Para una revisión más detallada de los impactos de la sequía, toda la región de estudio se dividió en los tres tipos de vegetación predominante (pastizales, matorrales y bosques) para examinar los cambios en Landsat NDWI y NDVI para ver las diferencias en la respuesta de las comu- nidades de plantas a la sequía severa. Los resultados confirmaron un mayor estrés por la sequía del 2013 y 2014 en los pastizales, en comparación con los matorrales y bosques en ambos años en todas las elevaciones y pendientes de más de 200 metros y 5%, respectivamente.
The year 2013 was the driest on record in gion meets the dry, warmer climate of the California, with a total precipitation of just southern region (Daly et al. 2008). The high- 30% of the statewide average (Hanak et al. est peaks in the Santa Lucia Mountains on 2014). The previous statewide record low in the Central Coast form a wall behind coastal 1976 was 56% of average precipitation. This hillsides, which effectively traps cooler extreme lack of rainfall during the 2013 and marine air (Henson and Usner 1996). This 2014 wet seasons created an unprecedented can lead to the formation of steep elevation level of stress on natural ecosystems and agri- gradients in air temperature regimes, humid- cultural production systems throughout the ity, and other climatic factors. This region of central regions of the state (Howitt et al. the state is continuously subjected to ex- 2014), where there remains a pressing need treme climatic events (Hiatt et al. 2012, Pot- to develop assessment methods for mitigating ter 2014a), frequent wildfires, and human local impacts and future risks. development impacts, together affecting wild- This drought assessment study focused on life populations, native plant communities, the Central Coast region, where the wet, soil conservation, riparian zones, and water cooler climate of the northern California re - quality.
1National Aeronautics and Space Administration, Ames Research Center, Moffett Field, CA 94035. E-mail: [email protected]
129 130 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Fig. 1. Central Coast study region within Monterey County and San Luis Obispo County, California. The dashed white outline delineates the general boundaries of the Big Sur subregion of approximately 140 km on the coast and 30 km inland.
The ecosystems of California’s Central the impacts of recent wildfires on vegetation Coast region have a high degree of biological growth (or regrowth) rates. diversity and endemism, and provide critical habitat for a large number of rare, endan- METHODS gered, and threatened animal and plant spe- Study Region cies (Henson and Usner 1996), including the wild orchid Piperia yadonii, the California California’s Central Coast region was Condor (Gymnogyps californianus), Califor- selected as the study region (Fig. 1). The Big nia southern sea otter (Enhydra lutris nereis), Sur subarea is generally delineated within this and the southernmost coast redwoods (Se - region by the 140 km (90 miles) of coastline quoia sempervirens). Mixed herbaceous–shrub from the Carmel River in Monterey County vegetation communities dominate the Central south to the San Carpoforo Creek in San Luis Coast region of California near Big Sur (Mon- Obispo County, extending about 30 km inland terey County). to the eastern foothills of the Santa Lucia The objective of this study was to assess the Mountains. The terrain is rugged and undulat- impacts of the historic 2013/14 drought on ing, with the steepest elevation gradients on ecosystems of the Central Coast region using a the U.S. Pacific coast (Shreve 1927). combination of satellite image analysis and in The region has a Mediterranean climate of situ measurements of soil moisture at sites warm, dry summers and cool, wet winters, representative of the predominant vegetation with localized summer fog near the coast types of the region. The novel methodology (Hiatt et al. 2012). Annual rainfall is highly used differencing of image data for the previ- variable and ranges from 40 cm to 155 cm ous above-average rainfall year in the region throughout the region, with highest event totals to successive years of far-below-average rain- normally falling on the higher mountains in fall years. The methodology also controlled for the northern extreme of the study area during 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 131 winter storms (Potter 2014a). During the sum- the Monterey Ranger District of the Los mer, fog and low clouds are frequent along the Padres National Forest (De Santis et al. 2010). coast. Mean annual temperature ranges be - During the recent drought period, the Pfeiffer tween 10 °C and 15 °C, from shaded canyon Ridge Fire burned more than 360 ha of forest bottoms to exposed ridgetops, respectively. over a 5-day period starting on 17 December Hiatt et al. (2012) reported that fog deposi- 2013. tion on land around Big Sur commonly peaks Climate Summary Data in July and August. This increase in fog depo- sition coincides with the period of lowest rain- Monthly climate summaries from 1950 to fall, characteristic of California’s Mediter- 2014 for the study area were obtained from ranean climate. Most summer fog deposition the Western Regional Climate Center data measured by Hiatt et al. (2012) at Big Sur portal (Abatzoglou et al. 2009, WRCC 2014). occurred during the night and early morning Two long-term data records from National hours. After 9:00, average fog deposition steeply Oceanic and Atmospheric Association Coopera- declined, to nearly zero between 14:00 and tive Stations were available for this study: Big 18:00. High wind speeds typically accompa- Sur State Park (BSP, National Weather Service nied increased rates of fog deposition during [NWS] ID 040790; 36°15 N, 121°47 W; eleva- periods of low dew-point depression. tion 320 m) and San Clemente Dam (SCD; Plant growth on the Central Coast can be NWS ID 047731; 36°26 N, 121°42 W; eleva- highly variable from year to year and is gener- tion 180 m). Locations of the stations are shown ally limited by declines of soil moisture in the in Fig. 1. summer and by cool temperatures in the win- The BSP record represents the longest ter (Potter 2014a, 2014b). The annual produc- complete climate record on the Monterey tion pattern for coastal grasses is rapid growth County coast south of Carmel, dating back to in the late fall (November) after the first rains 1950 for rainfall measurements. The station is have returned, slow winter growth (Decem- situated 3.5 km due east from the ocean at a ber–February), and rapid growth again in relatively low elevation. The SCD station is spring (March–May) (Shreve 1927, Corbin et located about 30 km inland in the Carmel al. 2005). Valley from BSP, on the eastern side of the Drier, southeast-facing slopes share a rela- Santa Lucia Mountains, in a location less in - tively equal distribution of coyote brush (Bac- fluenced by the direct on-shore flow of marine charis pilularis) and California coffeeberry air masses than is BSP. (Rhamnus californica) along with some Cali- Soil Moisture Monitoring fornia sagebrush (Artemisia californica) (Shreve 1927, Henson and Usner 1996). The coastal Volumetric water content (VWC) of grass- scrub community is usually a successional land, shrubland, and forest soils were moni- plant community that, in the absence of fire, tored from 2011 to 2014 at the U.S. Forest gradually moves into herbaceous (grassland) Service Brazil Ranch (36°24 N, 121°54 W) cover where the soil depth transitions from site (Fig. 2, Table 1), located approximately the shallowest to intermediate depth. The 12 km northwest of BSP (Potter 2014b). EC-5 herbaceous plant community includes Califor- frequency domain probes (Decagon Devices, nia annual grassland series and California oat- Pullman, WA) were placed at soil depths of grass series. Mixed evergreen forest stands 10 cm, 30 cm, and 50 cm for hourly measure- around Big Sur are dominated in the valley ments of VWC before and throughout the bottoms by coast redwoods and on steep slopes 2013–2014 drought period. The grassland and ridgetops by tanoaks, live oaks, madrones, soils on Serra Ridge were regularly grazed by manzanitas, pines, and firs (Henson and Usner cattle. The shrubland soils monitored were 1996). covered by coastal scrub vegetation on the Wildfire is a naturally occurring phenome- opposite slope of Serra Ridge. Forested soils non in Mediterranean ecosystems of the Cali- under coast redwood stands were monitored fornia Central Coast. In June and July of 2008, along Serra Creek canyon. Predominant soil 99,000 ha of the study region was burned by types on the Brazil Ranch included the the third-largest forest fire in California’s his- Sheridan, the Pfeiffer–Rock Outcrop com- tory, the Basin Complex and Indians Fire, on plex, and the Gamboa–Sur complex. Much of 132 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Fig. 2. Brazil Ranch image, with locations of soil moisture measurement stations.
TABLE 1. Site locations for soil moisture monitoring at the Brazil Ranch. Site number Vegetation cover Latitude Longitude Elevation (m) 1 Grassland 36.35556° N 121.89556° W 362 2 Shrubland 36.36139° N 121.89056° W 188 3 Forest 36.35833° N 121.88222° W 172 the Brazil Ranch property is covered by on near-peak seasonal patterns of plant growth coarse sandy loam soils on slopes >30% (Pot- or any dieback of canopy green cover. ter 2014b). All Landsat images used in this study were geometrically registered using terrain correc- Satellite Image Analysis tion algorithms (Level 1T) applied by the Cloud-free imagery from the Landsat 5 and USGS Earth Resources Observation and Sci- 8 Enhanced Thematic Mapper sensors was ence (EROS) Data Center (Sioux Falls, SD), selected from the U.S. Geological Survey and then converted to at-sensor reflectance (USGS) Earth Explorer data portal (USGS following the algorithms from Chander et al. 2014a). Predrought Landsat 5 image data from (2009). No further corrections for atmospheric path/row 43/35 were acquired on 3 May 2010, effects were applied, since the reflectance near the peak of the vegetation growing sea- indices used in this study employed near- son on the Central Coast (Potter 2014b). For infrared (NIR; 0.77–0.90 mm) and short-wave drought severity mapping, Landsat 8 images infrared (SWIR; 1.55–2.29 mm) wavelengths on 11 May 2013 and 14 May 2014 were also that are minimally affected by atmospheric acquired. By comparing May Landsat data for scattering (Avery and Berlin 1992, Gao 1996). all years, the methodology effectively controls The severity of drought stress at 30-m for seasonal temperature and solar radiation ground resolution across the study region was 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 133 determined using the Landsat normalized dif- indicate dense canopy vegetation cover. Nega- ference water index (NDWI; Gao 1996), tive NDVI values often indicate water bodies. which has been validated repeatedly in the Spatial Data Layers field as an index of vegetation moisture status (Hunt and Rock 1989, Chuvieco at al. 2002, Vegetation types for the study area were Dennison et al. 2005), as calculated in the fol- determined based on the National Agricul- lowing equation: tural Statistics Service California Cropland Data Layer (CDL) from 2012 (National Agri- NIR − SWIR cultural Statistics Service 2012). The CDL, NDWI = ______. NIR + SWIR which is a raster image, is a georeferenced, crop-specific land-cover data layer with a NDWI (scaled by 1000 to an integer for- ground resolution of 30 m. The CDL is pro- mat) is comparable to the Landsat normalized duced using satellite imagery from the Land- burn ratio used for wildfire severity mapping sat sensor collected during the current grow- (Miller and Yool 2002). ing season. Additional land-cover maps were The relative-differenced NDWI (RdNDWI) used as zonal layers for classifying results, was computed for this study as a compari- including the USGS National Elevation son of predrought and postdrought surface Dataset (USGS 2014b), the USGS National conditions: Land Cover Database 2001 (USGS 2001), and the National Aeronautics and Space Adminis- NDWI pre − NDWI post tration Moderate Resolution Imaging Spectro- RdNDWI = ______, √ABS (NDWI pre) radiometer (NASA 2014) 250-m resolution 16-day vegetation index composites. Based on where ABS indicates the absolute value of the the 2012 CDL layer, the 3 predominant vege- predrought NDWI; this value in the denomi- tation types in the study region (Fig. 3) were nator allowed computation of the square root grasslands (estimated to cover 517,370 ha), without changing the sign of the RdNDWI. shrublands (estimated to cover 556,350 ha), and High positive RdNDWI values would repre- forests (estimated to cover 146,110 ha). Crop- sent a decrease in vegetation moisture and land areas of the region defined by the CDL high drought stress or wildfire burn severity, were not included in this study because of the whereas near-zero values would represent rela- prevalence of irrigation to alleviate drought tively high vegetation moisture content and no stress and the frequency of crop type rotation in crease in drought stress. from year to year. The normalized difference vegetation index Perimeter boundaries for wildfires that oc - (NDVI) from Landsat imagery has been fre- curred within the study region since 2005 quently used to monitor live vegetation canopy were delineated from the database compiled cover and trajectories of change during and by the California Department of Forestry, Fire after ecosystem disturbances (Van Wagten- and Resource Assessment Program (2014). donk and Root 2003, Lentile et al. 2006). Burn severity classes at 30-m resolution for NDVI was computed for this study using the 2008 Basin Complex fire were determined Landsat bands: by the Monitoring Trends in Burn Severity (MTBS) project, which has consistently mapped NIR − Red the burn severity (from RdNBR methods) NDVI = ______, NIR + Red and perimeters of fires >405 ha (1000 acres) across the USA from 1984 to the present where Red is the reflectance from 0.63 m to (Eidenshink et al. 2007). The MTBS project is 0.69 m and NDVI was scaled from 0 to 1000. conducted through a partnership between the Advantages of NDVI for the purpose of vege- USGS EROS and the USDA Forest Service tation monitoring have been cited to include Remote Sensing Applications Center. Although its mathematical simplicity and ease of compa- there have been other large wildfires in the rability across numerous multispectral remote study region prior to 2005, only the most sensing platforms (Lentile et al. 2006). Low recent fires (since 2005) were examined, due NDVI values (near 0) indicate barren land to rapid postfire vegetation regrowth mea- cover whereas high NDVI values (near 1000) sured and verified by Li and Potter (2012). 134 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Fig. 3. Regional map layers for vegetation, elevation, slope, and aspect, all at a 30-m spatial resolution.
Elevation, slope, and aspect were deter- sets (Lehmann 2006). The K–S test does not mined at a 30-m spatial resolution (Fig. 3) assume that data were sampled from Gaussian from the USGS National Elevation Dataset distributions (or any other defined distribu- (USGS 2014b). For each 30-m cell, slope (as a tions), nor can its results be affected by chang- percentage) was calculated as the maximum ing data ranks or by numerical (e.g., logarith- rate of change in elevation value from that mic) transformations. The K–S test reports cell to its neighbors to determine the steepest the maximum difference between the 2 cumu- downhill descent from the cell. Aspect was lative distributions and calculates a P value calculated by fitting a plane to the z-values of from that difference and the sample sizes. It a 3 × 3 cell neighborhood around each 30-m tests the null hypothesis that both groups cell (Burrough and McDonell 1998). The di - were sampled from populations with identical rection that the plane faced was set as the distributions according to different medians, aspect for the cell. Aspect was expressed in variances, or outliers. If the K–S P value is degrees, moving clockwise from 0 (due north) small (i.e., <0.05), it can be concluded that to 360 (again due north). the 2 groups were sampled from populations with significantly different distributions. Statistical Analysis Tests of statistical significance between RESULTS sampled groups of points were carried out using the 2-sample Kolmogorov–Smirnov Regional Rainfall Deficits (K–S) test, a nonparametric method that com- For 2013, precipitation totaled 18.5 cm at pares the cumulative distributions of 2 data the BSP weather station, which was only 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 135
A BSP Weather Station
B SCD Weather Station
Fig. 4. Annual (and Jan–Apr monthly) precipitation from 1950 to 2014: A, Big Sur State Park (BSP) weather station; B, San Clemente Dam (SCD) weather station. about 17% of the long-term mean annual total the BSP station, which was 47% of long-term of 104 cm recorded at the station from 1950 to mean precipitation for the first half of the 2014 (Fig. 4A). Total precipitation during the years from 1950 to 2014. Similarly, for 2013 at first half (Jan–Jun) of 2014 was just 31.2 cm at the SCD weather station, precipitation totaled 136 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Grassland soil moisture
Shrubland soil moisture
Fig. 5. Caption on facing page. 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 137
Forest soil moisture
Fig. 5. Plots of hourly soil moisture from 2011 to 2014 at Brazil Ranch measurement stations: grassland, shrubland, and forest. to 11.6 cm, which was only about 21% of the Nacimiento lakes, were reported at 4% and 17% long-term mean annual total of 54 cm recorded of storage capacity, respectively. These storage at the station (Fig. 4B). Total precipitation dur- levels in 2014 were only about 8% and 47%, ing the first half (Jan–Jun) of 2014 was just respectively, of the storage levels at the same 23.4 cm at the SCD station, which was 65% of time period in 2012 (California Department of long-term mean precipitation for the first half Water Resources 2014). of the years from 1950 to 2014. For historical Soil Moisture Changes context, 2007, 1990, and 1976 were calendar years when <50% of the long-term average The record of hourly soil moisture for sites annual precipitation was recorded at both the at the Brazil Ranch from 2011 to late 2014 BSP and SCD weather stations. Calendar year (Fig. 5) showed that the months of May to 2010 was the last previous year with annual September 2013 had the lowest consistent precipitation above the long-term average. VWC at all 3 soil depths during the 3-year Reports of river flow rates and reservoir time series for these different vegetation cover storage confirmed the exceptional drought sta- types. The grassland site at Brazil Ranch was tus of the study region in 2014. As of mid- recorded every day of the month from May to August 2014, the Monterey Peninsula Water September 2013 and 2014 with hourly VWC Management District reported that the Carmel levels continuously below 20% at a soil depth River was no longer flowing out of the Ventana of 50 cm. The grassland soil VWC measured at Wilderness Area and into the Los Padres Dam a 10-cm depth was more variable, with rainfall reservoir. As of mid-September 2014, the 2 inputs from October to April of these years, large dammed reservoirs in the southern sec- but never exceeded 5% VWC from 1 March to tion of the study region, San Antonio and 22 September 2013. In contrast, the shrubland 138 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Fig. 6. Landsat 30-m normalized difference water index change maps for May 2013 and May 2014, differenced from May 2010 (predrought) for the study region. Wildfire perimeter boundaries from 2005 to 2012 are shown in gray out- lines. Boundaries of the Brazil Ranch and Hastings Natural History Reservation (Carmel Valley) are shown for geo- graphic reference locations in blue outlines.
TABLE 2. Landsat drought stress category values between May of 2010 and 2013/2014 for sites at the Brazil Ranch.
______RdNDWI ______dNDVI Site number Vegetation cover 2013 2014 2013 2014 1 Grassland 246 –24 –100 78 2 Shrubland 22 70 44 66 3 Forest –36 4 40 46 and the forest sites were re corded with hourly Ranch sites (Table 2) corresponded to the dif- soil VWC levels at or above 9% from May to ferences in VWC measured among the sites September in both 2013 and 2014 at a 50-cm during the period from July to September depth, and at higher than 5% hourly VWC at a 2013. Specifically, the grassland site was 10-cm depth during these drought periods detected with a RdNDWI value in May 2013 until June 2014. After June 2014, hourly soil of >200, whereas the shrubland and forest VWC levels at all 3 sites were at (or below) sites were detected with RdNDWI values <25 detection levels for the 10-cm and 30-cm depth in May 2013. The low soil moisture grassland sensors. response in 2013 implied that RdNDWI val- ues >200 (and differenced NDVI [dNDVI] Landsat Drought Assessment values lower than –100) were indicative of Change levels in Landsat NDWI and NDVI high-severity drought impacts on eco systems between May 2010 and May 2013 at the Brazil of the Central Coast region. 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 139
Fig. 7. Landsat 30-m normalized difference vegetation index change maps for May 2013 and May 2014, differenced from May 2010 (predrought) for the study region. Fire perimeter boundaries from 2005 to 2012 are shown in red out- lines. Boundaries of the Brazil Ranch and Hastings Natural History Reservation (Carmel Valley) are shown for geo- graphic reference locations in blue outlines.
It should be noted that the changes in into 3 predominant vegetation types (grass- Landsat NDWI and NDVI between May 2010 lands, shrublands, and forests) to examine and May 2014 at the Brazil Ranch sites were changes in Landsat NDWI and NDVI in the all well below these high-severity drought context of differing plant community responses thresholds detected in May 2013 (Table 2). to extreme drought. Results indicated that, by The lowered drought stress conditions at May 2013, 89% of all grasslands and 51% of Brazil Ranch sites likely resulted from late- all shrublands in the study region showed season precipitation events near Big Sur in relatively high drought stress, with Landsat April 2014, which totaled to about 6 cm of RdNDWI values >200. In contrast, only 4.7% rainfall at the BSP weather station. of all forest cover in the study region showed According to differences in Landsat NDWI this level of relatively high drought stress. and NDVI at the regional level between May Within each vegetation cover type, 1000 2010 and May 2013 (Figs. 6, 7), geographic points were randomly selected for statistical areas within the larger study region that were comparisons and were further characterized most severely impacted by the 2013 drought according to elevation, slope, aspect, and were the inland Carmel Valley in northern recent wildfire history. Results showed that a Monterey County, and the coast zones around significant difference (K–S test: P < 0.01) was San Simeon Point and Cambria in northern San detected between the frequency distribution Luis Obispo County. An expanded area of se - of grassland RdNDWI (mean RdNDWI > vere vegetation moisture stress, indicated by 440) and dNDVI (mean dNDVI < −170) for RdNDWI values >400, was detected by May both 2013 and 2104, compared to the fre- 2014 in both of these same geographic areas. quency distribution of RdNDWI and dNDVI For more detailed examination of drought of both shrublands and forests in the corre- impacts, the entire study region was separated sponding years, with grasslands showing more 140 WESTERN NORTH AMERICAN NATURALIST [Volume 75
TABLE 3. Comparison of Landsat drought stress category values between May of 2010 and 2013/2014 among the 3 predominant vegetation cover types in the Central Coast study region. Mean (SE) category values (for N = 1000 ran- domly selected locations) showed significant differences at P < 0.01 (Kolmogorov–Smirnov nonparametric test) among all 3 vegetation cover types in both drought stress categories.
______RdNDWI ______dNDVI Vegetation type 2013 2014 2013 2014 Grasslands 448 (5) 501 (4) –182 (1) –177 (1) Shrublands 200 (4) 236 (4) –76 (1) –80 (1) Forests –26 (2) 29 (2) 29 (1) 3 (1) severe drought stress overall (Table 3). Shrub- (N = 697) in 2014, whereas recently burned lands were also detected with more significant forest locations showed lowered drought stress, drought stress (mean RdNDWI > 200 and with mean RdNDWI = −152 (N = 303) in dNDVI < −70) overall than forest locations. 2014. Less than 2% of all grassland areas sam- Although no significant differences (at P < pled were categorized as recently burned, 0.05) could be detected in any of the 3 vegeta- which indicated that wildfire was a not a ma - tion types between 2013 and 2014 frequency jor factor influencing the drought response distributions of RdNDWI or dNDVI, mean of these herbaceous vegetation communities values of RdNDWI for 2014 were slightly across the study region. more extreme than for 2013 for all 3 vegeta- tion types. DISCUSSION Plots of sampled RdNDWI and dNDVI values by elevation and slope for each of the 3 Rainfall records from weather stations in predominant vegetation types (Figs. 8, 9, 10) the study region revealed that exceptional confirmed higher overall drought stress de- drought conditions were experienced in 2013 tected in grasslands compared to shrublands across the Central Coast area of California at and forests in both years at all elevations and 17%–21% of long-term average annual rainfall slopes >200 m and >5%, respectively. Drought totals compared to a statewide 30% of average stress in shrublands tended to peak at be - in 2013, compared to long-term rainfall totals tween 400 m and 500 m elevation in both years. (Hanak et al. 2014). In contrast to grasslands Drought stress in forests tended to be slightly and pasture lands on the Central Coast, by higher overall at all elevations below 1000 m May 2013 (prior to the Pfeiffer Ridge fire), and on slopes of <10%. There was no signifi- <5% of all forest cover area in the study cant variation by aspect in the Landsat drought region showed relatively high drought stress stress categories in either year for any of the (Landsat RdNDWI > 200). Within areas 3 vegetation types. burned by wildfire since 2005 across the study It was evident from a regional viewpoint region (made up of 90% forest and shrubland that changes in Landsat NDVI between May cover), <2% of all forest cover burned showed of 2010 and 2013/2014 were elevated in the relatively high drought stress. This was com- high-burn-severity areas of the 2008 Basin pared to 31% of all shrubland cover that showed Complex fire (labeled in Fig. 7) compared to relatively high drought stress (RdNDWI > surrounding unburned areas. This relationship 200) within burned areas. The overall conclu- of vegetation regrowth patterns in recent wild- sion from these results was that forested areas fire areas (2005–2012) to lowered vegetation in the study region were much less affected by drought stress was further confirmed for the moisture deficits in 2013 compared to shrub- sampled frequency distribution of shrubland lands, particularly within recently burned locations, with unburned locations showing areas where young trees recovering from wild- mean RdNDWI > 260 (N = 782) in both 2013 fire continued to show increases in green leaf and 2014, whereas recently burned shrubland cover and near-normal canopy moisture levels. locations showed much lower drought stress, The results from this Landsat image analy- with mean RdNDWI < −15 (N = 218) in sis showed that grasslands on the Central both 2013 and 2014. Similarly, unburned for- Coast were the ecosystems under the highest est locations showed mean RdNDWI = 170 drought stress in 2013 and 2014. Statewide 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 141
Grassland Cover
Fig. 8. Plots of Landsat relative-differenced normalized difference water index and differenced normalized difference vegetation index by elevation and slope, sampled at 1000 randomly selected locations within grassland cover across the study region. drought assessments have likewise reported Howitt et al. (2014) projected a 3% loss of rev- that dairy and cattle ranchers have been enue for the California livestock sector in 2014 among the first landowners to be impacted by due to lack of pasture and forage, and a $100 California water shortages, because livestock million reduction in statewide gross revenues. require forage year-round (Howitt et al. 2014). The native perennial grasses of the Central In contrast, valley crop growers typically have Coast region can survive droughts and then until mid-March to make planting decisions. regenerate during subsequent rainfall periods Statewide assessments have reported that (Hamilton et al. 2002). Small patches of peren- the lack of rainfall in the winter of 2013/14 nial grasses, often found at the periphery of substantially reduced the quality of nonirri- open herbaceous cover areas (Stromberg and gated pasture and the number of cattle per Griffin 1996), will persist in a nongreen state acre during the crucial periods of calving and during prolonged droughts. Landsat images in raising of feeder cattle on pasture. This lack of areas of high native grass and forb cover pasture feed caused ranchers to sell many cat- would thereby observe a high fraction of what tle out of the state, often to save irrigation appears to be barren soil in what are classified water for higher-value nonforage crops. The as grasslands from satellite mapping data. drought increased 2014 alfalfa hay prices in Satellite image analysis provides a high level the state by nearly 20% compared to 2013, of spatial detail in the assessment of drought making feedlot operations more expensive. impacts that no other regional monitoring 142 WESTERN NORTH AMERICAN NATURALIST [Volume 75
Shrubland Cover
Fig. 9. Plots of Landsat relative-differenced normalized difference water index and differenced normalized difference vegetation index by elevation and slope, sampled at 1000 randomly selected locations within shrubland cover across the study region. methodology can match. An important man- summer fog deposition in mitigating the lack agement use of the Landsat products gener- of winter rainfall in California coastal ecosys- ated in this study will be to estimate the tems. Hiatt et al. (2012) presented preliminary potential fire fuel loads of dry woodland and data to support the hypothesis that shrub shrubland vegetation biomass that will be sus- cover on the California coast intercepts mois- ceptible to burning, which could severely dam - ture from fog deposition that may help sustain age nearby residential and commercial build- the productivity of this vegetation community ings in the years to come. The effectiveness of through periods of low precipitation. Experi- wildfire fuel management treatments at many ments can be designed to determine the wind locations on the Big Sur coast can be readily and humidity conditions under which coastal assessed and compared during the next few shrub foliage may trap advected fog water and dry seasons by public land managers and pri- drip moisture into the surrounding soil below vate property owners. to mediate losses due to evapotranspiration. The results from this study will set the The second research effort will be to stage for at least 2 lines of ecological research develop effective methods to monitor changes on the California Central Coast region in the in vegetation habitats for rare and endangered (postdrought) years that will follow. The first species on the Big Sur coast. This type of post- research effort will be to investigate the role of drought plant species monitoring dates back 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 143
Forest Cover
Fig. 10. Plots of Landsat relative-differenced normalized difference water index and differenced normalized differ- ence vegetation index by elevation and slope, sampled at 1000 randomly selected locations within forest cover across the study region. to the mid-20th century (Weaver and Albert- upstream end of the gorge pool at the bound- son 1943). Today, the Central Coast serves as a ary of Pfeiffer Big Sur State Park with the Ven- refuge for biological species with a low toler- tana Wilderness Area. However, this section of ance for climate warming trends (Henson and the Big Sur River is above a fish barrier and is Usner 1996). If extreme summer temperatures not reachable by anadromous salmonids. There- observed in the coastal weather station rec- fore, in large part due to low stream flows dur- ords since the mid-1990s (Potter 2014a) con- ing the 2013/14 drought, the California Fish tinue as long-term climate change occurs far- and Game Commission has proposed to re- ther along the California coastal region, the move the current regulatory language autho- productivity and health of coastal rangelands rizing the harvest of hatchery trout and steel- and surrounding vegetation habitats may be head and return this section of the Big Sur altered in unanticipated ways. Changes in River to catch-and-release angling. growing season initiation date and duration for Growth of invasive plant species and patho - native plant species may result from climate gens can be altered during drought periods as warming. Future droughts may have large im - well. A prime example is sudden oak death pacts on river and steam flows on the Pacific (SOD), an exotic disease caused by the micro- coast. Under current regulations, the harvest scopic pathogen Phytophthora ramorum, intro- of hatchery trout and steelhead is allowed on duced to California 20–25 years ago. Manag- the Big Sur River and tributaries above the ing the presence of California bay laurel 144 WESTERN NORTH AMERICAN NATURALIST [Volume 75
(Umbellularia californica), the primary carrier CHANDER, G., B. MARKHAM, AND D. HELDER. 2009. Sum- of P. ramorum in California, around high-value mary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sen- native oak stands in the Big Sur region may sors. Remote Sensing of the Environment 113: reduce the number of infectious SOD spores. 893–903. In 2014, only 8 of 170 bay laurel trees sur- CHUVIECO, E., D. RIANO, I. AGUADO, AND D. COCERO. veyed near the Post Ranch in Big Sur tested 2002. Estimation of fuel moisture content from multi- temporal analysis of Landsat Thematic Mapper re- positive for SOD disease; only 1 tree out of flectance data: application in fire danger assess- 104 tested at Pfeiffer Big Sur State Park tested ment. International Journal of Remote Sensing 23: positive for SOD (Garbelotto 2014). During 2145–2162. the recent period of relatively low SOD spread, CORBIN, J.D., M.A. THOMSEN, T.E. DAWSON, AND C.M. selective removal and pruning of drought- D’ANTONIO. 2005. Summer water use by California coastal prairie grasses: fog, drought, and community stressed bay laurel trees may be an effective composition. Oecologia 145:511–521. strategy to protect native oaks from further DALY, C., M. HALBLEIB, J.I. SMITH, W.P. GIBSON, M.K. SOD infection. DOGGETT, G.H. TAYLOR, J. CURTIS, AND P.A. PAS- In summary, sudden shifts in climate and TERIS. 2008. Physiographically sensitive mapping of temperature and precipitation across the contermi- the frequency of extreme weather events, such nous United States. International Journal of Clima- as the 2013/14 drought, have the potential to tip tology 28:2031–2064. the balance in favor of or against vulnerable DENNISON, P.E., D.A. ROBERTS, S.H. PETERSON, AND J. biological resources and the human activities RECHEL. 2005. Use of normalized difference water that sometimes require a struggle to prosper index for monitoring live fuel moisture. Interna- tional Journal of Remote Sensing 26:1035–1042. in a remote and rugged coastal environment DE SANTIS, A., G.P. ASNER, P.J. VAUGHAN, AND D.E. KNAPP. such as the California Central Coast. Reliable, 2010. Mapping burn severity and burning efficiency low-cost monitoring methods of the regional in California using simulation models and Landsat land cover are an essential part of the ongoing imagery. Remote Sensing of Environment 114: effort to prepare for future droughts. 1535–1545. EIDENSHINK J., B. SCHWIND, K. BREWER, Z. ZHU, B. QUAYLE, AND S. HOWARD. 2007. A project for moni- ACKNOWLEDGMENTS toring trends in burn severity. Fire Ecology Special Issue 3:3–21. The author acknowledges the cooperation GAO, B.C. 1996. NDWI—a normalized difference water index for remote sensing of vegetation liquid water of the U.S. Forest Service for access to the from space. Remote Sensing of Environment 58: Brazil Ranch property and for collaboration on 257–266. climate-ecosystem science investigations in GARBELOTTO, M. 2014. UC Berkeley Forest Pathology and Big Sur. Thanks also to Chris Hauser and Mycology Lab. http://nature.berkeley.edu/garbelot Leslie Dorrance of the Santa Lucia Conser- towp/ HAMILTON, J.G., J.R. GRIFFIN, AND M.R. STROMBERG. vancy and Sarah Hardgrave of the Big Sur 2002. Long-term population dynamics of native Nas- Land Trust for feedback on the potential sup- sella bunchgrasses in unmanaged stands in central port for land management decisions from California. Madroño 49:274–284. satellite image analysis. HANAK, E., J. MOUNT, AND C. CHAPPELLE. 2014. Califor- nia’s latest drought. Public Policy Institute of Cali- fornia February 2014. http://www.ppic.org/main/ LITERATURE CITED publication_show.asp?i=1087 HENSON, P., AND D.J. USNER. 1996. The natural history of ABATZOGLOU, J.T., K.T. REDMOND, AND L.M. EDWARDS. Big Sur. University of California Press, Berkeley, 2009. Classification of regional climate variability in CA. 416 pp. the state of California. Journal of Applied Meteorol- HIATT, C., D. FERNANDEZ, AND C. POTTER. 2012, Mea- ogy and Climatology 48:1527–1541. surements of fog water deposition on the California AVERY, T.E., AND G.L. BERLIN. 1992. Fundamentals of Central Coast. Atmospheric and Climate Sciences remote sensing and airphoto interpretation. Prentice 2:525–531. Hall, Upper Saddle River, NJ. 472 pp. HOWITT, R.E., J. MEDELLIN-AZUARA, D. MACEWAN, J.R. BURROUGH, P.A., AND R.A. MCDONELL. 1998. Principles LUND, AND D.A. SUMNER. 2014. Economic analysis of geographical information systems. Oxford Univer- of the 2014 drought for California agriculture. Cen- sity Press, New York, NY. 190 pp. ter for Watershed Sciences, University of California, CALIFORNIA DEPARTMENT OF FORESTRY, FIRE AND Davis, CA. RESOURCE ASSESSMENT PROGRAM. 2014. Fire and HUNT, E.R, AND B.N. ROCK. 1989. Detection of changes in Resource Assessment Program [web application]. leaf water content using near and middle-infrared http://frap.cdf.ca.gov/ reflectances. Remote Sensing of Environment 30: CALIFORNIA DEPARTMENT OF WATER RESOURCES. 2014. 43–54. California Data Exchange Center—Reservoirs. LEHMANN, E. 2006. Nonparametrics: statistical methods http://cdec.water.ca.gov/reservoir.html based on ranks. Springer, New York, NY. 464 pp. 2015] EFFECTS OF DROUGHT ON COASTAL CALIFORNIA 145
LENTILE, L., Z.A. HOLDEN, A. SMITH, M. FALKOWSKI, A. STROMBERG, M.R., AND J.R. GRIFFIN. 1996. Long-term HUDAK, P. MORGAN, S. LEWIS, P. GESSLER, AND N. patterns in coastal California grasslands in relation to BENSON. 2006. Remote sensing techniques to assess cultivation, gophers and grazing. Ecological Applica- active fire characteristics and post-fire effects. Inter- tions 6:1189–1211. national Journal of Wildland Fire 15:319–345. [USGS] U.S. GEOLOGICAL SURVEY. 2001. National Land LI, S., AND C.S. POTTER. 2012. Patterns of aboveground Cover Database 2001 [web application]. http://www biomass regeneration in post-fire coastal scrub com- .mrlc.gov/nlcd2001.php munities. GIScience and Remote Sensing 49:182–201. [USGS] U.S. GEOLOGICAL SURVEY. 2014a. Earth Explorer MILLER, J.D., AND S.R. YOOL. 2002. Mapping forest post- [web application]. http://earthexplorer.usgs.gov/ fire canopy consumption in several overstory types [USGS] U.S. GEOLOGICAL SURVEY. 2014b. National Eleva- using multi-temporal Landsat TM and ETM data. tion Dataset [web application]. http://ned.usgs.gov/ Remote Sensing of Environment 82:481–496. VAN WAGTENDONK, J.W., AND R.R. ROOT. 2003. The use of [NASA] NATIONAL AERONAUTICS AND SPACE ADMINISTRA- multi-temporal Landsat Normalized Difference Vege- TION. 2014. MODIS [web application]. http://modis tation Index (NDVI) data for mapping fuel models in .gsfc.nasa.gov/ Yosemite National Park, USA. International Journal NATIONAL AGRICULTURAL STATISTICS SERVICE. 2012. Crop- Scape—Cropland Data Layer. http://nassgeodata of Remote Sensing 24(8):1639–1651. .gmu.edu/CropScape WEAVER, J.E., AND F.W. ALBERTSON. 1943. Resurvey of POTTER, C. 2014a. Understanding climate change on the grasses, forbs and underground plant parts at the California coast: accounting for extreme daily events end of the great drought. Ecological Monographs among long-term trends. Climate 2:18–27. 13:64–117. POTTER, C. 2014b. Microclimate influences on vegetation [WRCC] WESTERN REGIONAL CLIMATE CENTER. 2014. water availability and net primary production in Cooperative Climatological Data Summaries. http:// coastal ecosystems of Central California. Landscape www.wrcc.dri.edu/climatedata/climsum Ecology 29:677–687. SHREVE, F. 1927. The vegetation of a coastal mountain Received 1 December 2014 range. Ecology 8:27–44. Accepted 20 March 2015